natural 0.2.0 → 0.2.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.editorconfig +7 -0
- package/README.md +3 -3
- package/lib/natural/inflectors/present_verb_inflector.js +1 -0
- package/lib/natural/phonetics/double_metaphone.js +9 -4
- package/lib/natural/phonetics/metaphone.js +4 -0
- package/lib/natural/phonetics/soundex.js +11 -7
- package/lib/natural/stemmers/porter_stemmer.js +90 -62
- package/lib/natural/stemmers/porter_stemmer_pt.js +302 -0
- package/lib/natural/stemmers/stemmer_pt.js +65 -0
- package/lib/natural/stemmers/token.js +163 -0
- package/lib/natural/util/stopwords_pt.js +136 -0
- package/package.json +2 -2
- package/spec/double_metaphone_spec.js +10 -1
- package/spec/metaphone_spec.js +5 -0
- package/spec/porter_stemmer_pt_spec.js +50 -0
- package/spec/porter_stemmer_spec.js +45 -0
- package/spec/present_verb_inflector_spec.js +3 -0
- package/spec/soundex_spec.js +13 -9
- package/spec/stemmer_token_spec.js +161 -0
- package/spec/test_data/snowball_pt.txt +32016 -0
- package/coverage/coverage.json +0 -1
- package/coverage/lcov-report/index.html +0 -584
- package/coverage/lcov-report/io_spec/index.html +0 -350
- package/coverage/lcov-report/io_spec/tfidf_spec.js.html +0 -469
- package/coverage/lcov-report/lib/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/lib/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/lib/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/lib/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/lib/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/lib/natural/distance/index.html +0 -376
- package/coverage/lcov-report/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/lib/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/lib/natural/index.html +0 -350
- package/coverage/lcov-report/lib/natural/index.js.html +0 -547
- package/coverage/lcov-report/lib/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/lib/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/lib/natural/inflectors/fr/count_inflector.js.html +0 -520
- package/coverage/lcov-report/lib/natural/inflectors/fr/index.html +0 -363
- package/coverage/lcov-report/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/lib/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/lib/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/lib/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/lib/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/lib/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/lib/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/lib/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/lib/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/lib/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/lib/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/lib/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/lib/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/lib/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/lib/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/lib/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/lib/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/lib/natural/spellcheck/index.html +0 -350
- package/coverage/lcov-report/lib/natural/spellcheck/spellcheck.js.html +0 -562
- package/coverage/lcov-report/lib/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/lib/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/lib/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/lib/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/lib/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/lib/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/lib/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/lib/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/lib/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/lib/natural/trie/index.html +0 -350
- package/coverage/lcov-report/lib/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/lib/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/lib/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/lib/natural/util/index.html +0 -532
- package/coverage/lcov-report/lib/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/lib/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/lib/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/lib/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/lib/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/lib/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/lib/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/lib/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/lib/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/lib/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/lib/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/lib/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/lib/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/lib/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/lib/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/lib/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/natural/distance/index.html +0 -376
- package/coverage/lcov-report/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/natural/index.html +0 -350
- package/coverage/lcov-report/natural/index.js.html +0 -433
- package/coverage/lcov-report/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/natural/inflectors/fr/index.html +0 -350
- package/coverage/lcov-report/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/natural/lib/natural/analyzers/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/analyzers/sentence_analyzer.js.html +0 -850
- package/coverage/lcov-report/natural/lib/natural/classifiers/bayes_classifier.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/classifiers/classifier.js.html +0 -823
- package/coverage/lcov-report/natural/lib/natural/classifiers/index.html +0 -376
- package/coverage/lcov-report/natural/lib/natural/classifiers/logistic_regression_classifier.js.html +0 -502
- package/coverage/lcov-report/natural/lib/natural/distance/dice_coefficient.js.html +0 -529
- package/coverage/lcov-report/natural/lib/natural/distance/index.html +0 -376
- package/coverage/lcov-report/natural/lib/natural/distance/jaro-winkler_distance.js.html +0 -670
- package/coverage/lcov-report/natural/lib/natural/distance/levenshtein_distance.js.html +0 -526
- package/coverage/lcov-report/natural/lib/natural/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/index.js.html +0 -547
- package/coverage/lcov-report/natural/lib/natural/inflectors/count_inflector.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/inflectors/form_set.js.html +0 -406
- package/coverage/lcov-report/natural/lib/natural/inflectors/fr/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/inflectors/fr/noun_inflector.js.html +0 -1027
- package/coverage/lcov-report/natural/lib/natural/inflectors/index.html +0 -402
- package/coverage/lcov-report/natural/lib/natural/inflectors/ja/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/inflectors/ja/noun_inflector.js.html +0 -727
- package/coverage/lcov-report/natural/lib/natural/inflectors/noun_inflector.js.html +0 -637
- package/coverage/lcov-report/natural/lib/natural/inflectors/present_verb_inflector.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/inflectors/singular_plural_inflector.js.html +0 -649
- package/coverage/lcov-report/natural/lib/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/natural/lib/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/natural/lib/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/natural/lib/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/natural/lib/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/natural/lib/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/natural/lib/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/natural/lib/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/natural/lib/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/natural/lib/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/natural/lib/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/natural/lib/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/natural/lib/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/natural/lib/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/natural/lib/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/natural/lib/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/natural/lib/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/natural/lib/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/natural/lib/natural/trie/index.html +0 -350
- package/coverage/lcov-report/natural/lib/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/natural/lib/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/natural/lib/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/natural/lib/natural/util/index.html +0 -532
- package/coverage/lcov-report/natural/lib/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/natural/lib/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/natural/lib/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/natural/lib/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/natural/lib/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/natural/lib/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/natural/lib/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/natural/lib/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/natural/lib/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/natural/lib/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/natural/ngrams/index.html +0 -363
- package/coverage/lcov-report/natural/ngrams/ngrams.js.html +0 -595
- package/coverage/lcov-report/natural/ngrams/ngrams_zh.js.html +0 -571
- package/coverage/lcov-report/natural/normalizers/index.html +0 -389
- package/coverage/lcov-report/natural/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/natural/normalizers/normalizer_ja.js.html +0 -2194
- package/coverage/lcov-report/natural/normalizers/normalizer_no.js.html +0 -496
- package/coverage/lcov-report/natural/normalizers/remove_diacritics.js.html +0 -685
- package/coverage/lcov-report/natural/phonetics/dm_soundex.js.html +0 -1063
- package/coverage/lcov-report/natural/phonetics/double_metaphone.js.html +0 -1840
- package/coverage/lcov-report/natural/phonetics/index.html +0 -402
- package/coverage/lcov-report/natural/phonetics/metaphone.js.html +0 -886
- package/coverage/lcov-report/natural/phonetics/phonetic.js.html +0 -484
- package/coverage/lcov-report/natural/phonetics/soundex.js.html +0 -562
- package/coverage/lcov-report/natural/stemmers/index.html +0 -571
- package/coverage/lcov-report/natural/stemmers/lancaster_rules.js.html +0 -2692
- package/coverage/lcov-report/natural/stemmers/lancaster_stemmer.js.html +0 -544
- package/coverage/lcov-report/natural/stemmers/porter_stemmer.js.html +0 -937
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_es.js.html +0 -985
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_fa.js.html +0 -418
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_fr.js.html +0 -1450
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_it.js.html +0 -1018
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_no.js.html +0 -784
- package/coverage/lcov-report/natural/stemmers/porter_stemmer_ru.js.html +0 -775
- package/coverage/lcov-report/natural/stemmers/stemmer.js.html +0 -505
- package/coverage/lcov-report/natural/stemmers/stemmer_es.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_fa.js.html +0 -484
- package/coverage/lcov-report/natural/stemmers/stemmer_fr.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_it.js.html +0 -427
- package/coverage/lcov-report/natural/stemmers/stemmer_ja.js.html +0 -736
- package/coverage/lcov-report/natural/stemmers/stemmer_no.js.html +0 -505
- package/coverage/lcov-report/natural/stemmers/stemmer_pl.js.html +0 -496
- package/coverage/lcov-report/natural/stemmers/stemmer_ru.js.html +0 -496
- package/coverage/lcov-report/natural/tfidf/index.html +0 -350
- package/coverage/lcov-report/natural/tfidf/tfidf.js.html +0 -904
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_es.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fa.js.html +0 -466
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_fr.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_it.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_nl.js.html +0 -430
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_no.js.html +0 -439
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pl.js.html +0 -457
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_pt.js.html +0 -442
- package/coverage/lcov-report/natural/tokenizers/aggressive_tokenizer_ru.js.html +0 -457
- package/coverage/lcov-report/natural/tokenizers/index.html +0 -519
- package/coverage/lcov-report/natural/tokenizers/regexp_tokenizer.js.html +0 -580
- package/coverage/lcov-report/natural/tokenizers/tokenizer.js.html +0 -475
- package/coverage/lcov-report/natural/tokenizers/tokenizer_ja.js.html +0 -1153
- package/coverage/lcov-report/natural/tokenizers/treebank_word_tokenizer.js.html +0 -544
- package/coverage/lcov-report/natural/transliterators/ja/index.html +0 -350
- package/coverage/lcov-report/natural/transliterators/ja/index.js.html +0 -2068
- package/coverage/lcov-report/natural/trie/index.html +0 -350
- package/coverage/lcov-report/natural/trie/trie.js.html +0 -1012
- package/coverage/lcov-report/natural/util/bag.js.html +0 -472
- package/coverage/lcov-report/natural/util/edge_weighted_digraph.js.html +0 -658
- package/coverage/lcov-report/natural/util/index.html +0 -532
- package/coverage/lcov-report/natural/util/longest_path_tree.js.html +0 -619
- package/coverage/lcov-report/natural/util/shortest_path_tree.js.html +0 -622
- package/coverage/lcov-report/natural/util/stopwords.js.html +0 -445
- package/coverage/lcov-report/natural/util/stopwords_es.js.html +0 -430
- package/coverage/lcov-report/natural/util/stopwords_fa.js.html +0 -436
- package/coverage/lcov-report/natural/util/stopwords_fr.js.html +0 -904
- package/coverage/lcov-report/natural/util/stopwords_it.js.html +0 -478
- package/coverage/lcov-report/natural/util/stopwords_ja.js.html +0 -499
- package/coverage/lcov-report/natural/util/stopwords_no.js.html +0 -445
- package/coverage/lcov-report/natural/util/stopwords_pl.js.html +0 -508
- package/coverage/lcov-report/natural/util/stopwords_ru.js.html +0 -445
- package/coverage/lcov-report/natural/util/topological.js.html +0 -616
- package/coverage/lcov-report/natural/util/utils.js.html +0 -679
- package/coverage/lcov-report/natural/wordnet/data_file.js.html +0 -580
- package/coverage/lcov-report/natural/wordnet/index.html +0 -389
- package/coverage/lcov-report/natural/wordnet/index_file.js.html +0 -742
- package/coverage/lcov-report/natural/wordnet/wordnet.js.html +0 -829
- package/coverage/lcov-report/natural/wordnet/wordnet_file.js.html +0 -529
- package/coverage/lcov-report/normalizers/index.html +0 -350
- package/coverage/lcov-report/normalizers/normalizer.js.html +0 -607
- package/coverage/lcov-report/prettify.css +0 -1
- package/coverage/lcov-report/prettify.js +0 -1
- package/coverage/lcov-report/spec/aggressive_tokenizer_es_spec.js.html +0 -430
- package/coverage/lcov-report/spec/aggressive_tokenizer_fr_spec.js.html +0 -625
- package/coverage/lcov-report/spec/aggressive_tokenizer_nl_spec.js.html +0 -430
- package/coverage/lcov-report/spec/aggressive_tokenizer_pt_spec.js.html +0 -490
- package/coverage/lcov-report/spec/aggressive_tokenizer_spec.js.html +0 -490
- package/coverage/lcov-report/spec/bayes_classifier_spec.js.html +0 -841
- package/coverage/lcov-report/spec/classifier_spec.js.html +0 -847
- package/coverage/lcov-report/spec/count_inflector_fr_spec.js.html +0 -454
- package/coverage/lcov-report/spec/count_inflector_spec.js.html +0 -538
- package/coverage/lcov-report/spec/dice_coefficient_spec.js.html +0 -400
- package/coverage/lcov-report/spec/double_metaphone_spec.js.html +0 -2548
- package/coverage/lcov-report/spec/index.html +0 -883
- package/coverage/lcov-report/spec/jaro-winkler_spec.js.html +0 -481
- package/coverage/lcov-report/spec/lancaster_stemmer_spec.js.html +0 -541
- package/coverage/lcov-report/spec/levenshtein_spec.js.html +0 -520
- package/coverage/lcov-report/spec/logistic_regression_classifier_spec.js.html +0 -688
- package/coverage/lcov-report/spec/longest_path_tree_spec.js.html +0 -628
- package/coverage/lcov-report/spec/metaphone_spec.js.html +0 -1063
- package/coverage/lcov-report/spec/ngram_spec.js.html +0 -712
- package/coverage/lcov-report/spec/ngram_zh_spec.js.html +0 -688
- package/coverage/lcov-report/spec/normalizer_ja_spec.js.html +0 -802
- package/coverage/lcov-report/spec/normalizer_no_spec.js.html +0 -655
- package/coverage/lcov-report/spec/normalizer_spec.js.html +0 -742
- package/coverage/lcov-report/spec/noun_inflector_fr_spec.js.html +0 -1750
- package/coverage/lcov-report/spec/noun_inflector_ja_spec.js.html +0 -658
- package/coverage/lcov-report/spec/noun_inflector_spec.js.html +0 -1132
- package/coverage/lcov-report/spec/porter_stemmer_fr_spec.js.html +0 -625
- package/coverage/lcov-report/spec/porter_stemmer_it_spec.js.html +0 -454
- package/coverage/lcov-report/spec/porter_stemmer_no_spec.js.html +0 -751
- package/coverage/lcov-report/spec/porter_stemmer_ru_spec.js.html +0 -511
- package/coverage/lcov-report/spec/porter_stemmer_spec.js.html +0 -733
- package/coverage/lcov-report/spec/present_verb_inflector_spec.js.html +0 -697
- package/coverage/lcov-report/spec/remove_diacritics_spec.js.html +0 -415
- package/coverage/lcov-report/spec/sentence_analyzer_spec.js.html +0 -1153
- package/coverage/lcov-report/spec/shortest_path_tree_spec.js.html +0 -718
- package/coverage/lcov-report/spec/soundex_spec.js.html +0 -784
- package/coverage/lcov-report/spec/spellcheck_spec.js.html +0 -598
- package/coverage/lcov-report/spec/stemmer_ja_spec.js.html +0 -520
- package/coverage/lcov-report/spec/tfidf_spec.js.html +0 -1288
- package/coverage/lcov-report/spec/tokenizer_ja_spec.js.html +0 -496
- package/coverage/lcov-report/spec/transliterator_ja_spec.js.html +0 -631
- package/coverage/lcov-report/spec/treebank_word_tokenizer_spec.js.html +0 -430
- package/coverage/lcov-report/spec/trie_spec.js.html +0 -940
- package/coverage/lcov-report/util/index.html +0 -350
- package/coverage/lcov-report/util/utils.js.html +0 -679
- package/coverage/lcov.info +0 -11094
- package/lib/natural/parsers/Agenda.js +0 -68
- package/lib/natural/parsers/CYKParser.js +0 -87
- package/lib/natural/parsers/CYK_Item.js +0 -159
- package/lib/natural/parsers/Chart.js +0 -211
- package/lib/natural/parsers/ChartParser.js +0 -60
- package/lib/natural/parsers/DoubleDottedItem.js +0 -226
- package/lib/natural/parsers/EarleyItem.js +0 -155
- package/lib/natural/parsers/EarleyParser.js +0 -58
- package/lib/natural/parsers/FunctionWordTagger.js +0 -84
- package/lib/natural/parsers/GoalItem.js +0 -206
- package/lib/natural/parsers/GrammarParser.js +0 -1289
- package/lib/natural/parsers/HeadCornerParser.js +0 -101
- package/lib/natural/parsers/LeftCornerParser.js +0 -58
- package/lib/natural/parsers/PEG-grammar-for-unification-grammar.txt +0 -345
|
@@ -1,206 +0,0 @@
|
|
|
1
|
-
/*
|
|
2
|
-
GoalItem class
|
|
3
|
-
Copyright (C) 2014 Hugo W.L. ter Doest
|
|
4
|
-
|
|
5
|
-
This program is free software: you can redistribute it and/or modify
|
|
6
|
-
it under the terms of the GNU General Public License as published by
|
|
7
|
-
the Free Software Foundation, either version 3 of the License, or
|
|
8
|
-
(at your option) any later version.
|
|
9
|
-
|
|
10
|
-
This program is distributed in the hope that it will be useful,
|
|
11
|
-
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
12
|
-
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
13
|
-
GNU General Public License for more details.
|
|
14
|
-
|
|
15
|
-
You should have received a copy of the GNU General Public License
|
|
16
|
-
along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
17
|
-
*/
|
|
18
|
-
|
|
19
|
-
var typeOf = require('typeof');
|
|
20
|
-
var log4js = require('log4js');
|
|
21
|
-
var logger = log4js.getLogger();
|
|
22
|
-
logger.setLevel('ERROR');
|
|
23
|
-
|
|
24
|
-
function GoalItem(i, j, X) {
|
|
25
|
-
logger.debug("Enter GoalItem: " + i + ", " + j + ", " + X);
|
|
26
|
-
this.id = "Goal(" + i + ", " + j + ", " + X + ")";
|
|
27
|
-
this.name = X;
|
|
28
|
-
this.children = [];
|
|
29
|
-
this.data = {};
|
|
30
|
-
this.data.from = i;
|
|
31
|
-
this.data.to = j;
|
|
32
|
-
this.data.nonterminal = X;
|
|
33
|
-
}
|
|
34
|
-
|
|
35
|
-
module.exports = GoalItem;
|
|
36
|
-
|
|
37
|
-
var DoubleDottedItem = require('./DoubleDottedItem');
|
|
38
|
-
var CYK_Item = require('./CYK_Item');
|
|
39
|
-
|
|
40
|
-
// Returns false
|
|
41
|
-
GoalItem.prototype.is_complete = function() {
|
|
42
|
-
logger.debug("Enter GoalItem.is_complete: a goal item always is incomplete!");
|
|
43
|
-
return(false);
|
|
44
|
-
};
|
|
45
|
-
|
|
46
|
-
GoalItem.prototype.combine_with_chart = function(chart, agenda, grammar) {
|
|
47
|
-
var nr_items_added = 0;
|
|
48
|
-
|
|
49
|
-
logger.debug("Enter GoalItem.combine_with_chart:" + this.id);
|
|
50
|
-
nr_items_added += this.hc_predict(chart, agenda, grammar);
|
|
51
|
-
nr_items_added += this.hc_predict_empty(chart, agenda, grammar);
|
|
52
|
-
nr_items_added += this.left_predict(chart, agenda, grammar);
|
|
53
|
-
nr_items_added += this.right_predict(chart, agenda, grammar);
|
|
54
|
-
nr_items_added += this.left_complete(chart, agenda, grammar);
|
|
55
|
-
nr_items_added += this.right_complete(chart, agenda, grammar);
|
|
56
|
-
logger.debug("Exit GoalItem.combine_with_chart: number of items added: " + nr_items_added);
|
|
57
|
-
return(nr_items_added);
|
|
58
|
-
};
|
|
59
|
-
|
|
60
|
-
// Adds double dotted items to the chart based on goal and CYK items
|
|
61
|
-
GoalItem.prototype.hc_predict = function(chart, agenda, grammar) {
|
|
62
|
-
var that = this;
|
|
63
|
-
var nr_items_added = 0;
|
|
64
|
-
|
|
65
|
-
logger.debug("Enter GoalItem.hc_predict: " + this.id);
|
|
66
|
-
var items = chart.get_items_from_to(this.data.from, this.data.to);
|
|
67
|
-
items.forEach(function(item) {
|
|
68
|
-
if (typeOf(item) === "cyk_item") {
|
|
69
|
-
// Create head-corner items
|
|
70
|
-
var rules = grammar.get_rules_with_head(item.data.rule.lhs);
|
|
71
|
-
rules.forEach(function(rule) {
|
|
72
|
-
if (grammar.is_headcorner_of(rule.lhs, that.data.nonterminal)) {
|
|
73
|
-
var new_item = new DoubleDottedItem(rule, rule.head, rule.head+1, that.data.from, that.data.to);
|
|
74
|
-
new_item.children.push(item);
|
|
75
|
-
nr_items_added += agenda.add_item(new_item, chart);
|
|
76
|
-
logger.debug("GoalItem.hc_predict: " + that.id +", " + item.id + " |- " + new_item.id);
|
|
77
|
-
}
|
|
78
|
-
});
|
|
79
|
-
}
|
|
80
|
-
});
|
|
81
|
-
logger.debug("Exit GoalItem.hc_predict: number of items added: " + nr_items_added);
|
|
82
|
-
return(nr_items_added);
|
|
83
|
-
};
|
|
84
|
-
|
|
85
|
-
// Add CYK items for epsilon rules with a left-hand-side that is head-corner of the goal
|
|
86
|
-
GoalItem.prototype.hc_predict_empty = function(chart, agenda, grammar) {
|
|
87
|
-
var that = this;
|
|
88
|
-
var nr_items_added = 0;
|
|
89
|
-
|
|
90
|
-
logger.debug("Enter GoalItem.hc_predict_empty: " + this.id);
|
|
91
|
-
if (this.data.from === this.data.to) {
|
|
92
|
-
grammar.production_rules.forEach(function(rule) {
|
|
93
|
-
if ((rule.rhs.length === 0) && grammar_is_headcorner_of(rule.lhs, that.data.nonterminal)) {
|
|
94
|
-
var new_item = new CYK_item(rule, that.data.from, that.data.to);
|
|
95
|
-
nr_items_added += agenda.add_item(new_item, chart);
|
|
96
|
-
logger.debug("GoalItem.hc_predict_empty: " + that.id + " |- " + new_item.id);
|
|
97
|
-
}
|
|
98
|
-
});
|
|
99
|
-
}
|
|
100
|
-
logger.debug("Exit GoalItem.hc_predict_empty: number of items added: " + nr_items_added);
|
|
101
|
-
return(nr_items_added);
|
|
102
|
-
};
|
|
103
|
-
|
|
104
|
-
GoalItem.prototype.left_predict = function (chart, agenda, grammar) {
|
|
105
|
-
var that = this;
|
|
106
|
-
var nr_items_added = 0;
|
|
107
|
-
|
|
108
|
-
logger.debug("Enter GoalItem.left_predict: " + this.id);
|
|
109
|
-
var items = chart.get_items_to(this.data.to);
|
|
110
|
-
items.forEach(function(item) {
|
|
111
|
-
if ((typeOf(item) === "doubledotteditem") && (item.data.left_dot > 0)){
|
|
112
|
-
if (grammar.is_headcorner_of(item.data.rule.lhs, that.data.nonterminal)) {
|
|
113
|
-
for (var i = that.data.from; i <= item.data.from; i++) {
|
|
114
|
-
for (var j = i; j <= item.data.from; j++) {
|
|
115
|
-
var new_goal = new GoalItem(i, j, item.data.rule.rhs[item.data.left_dot-1]);
|
|
116
|
-
nr_items_added += agenda.add_item(new_goal, chart);
|
|
117
|
-
logger.debug("GoalItem.left_predict: " + that.id +", " + item.id + " |- " + new_goal.id);
|
|
118
|
-
}
|
|
119
|
-
}
|
|
120
|
-
}
|
|
121
|
-
}
|
|
122
|
-
});
|
|
123
|
-
logger.debug("Exit GoalItem.left_predict: number of items added: " + nr_items_added);
|
|
124
|
-
return(nr_items_added);
|
|
125
|
-
};
|
|
126
|
-
|
|
127
|
-
GoalItem.prototype.right_predict = function (chart, agenda, grammar) {
|
|
128
|
-
var that = this;
|
|
129
|
-
var nr_items_added = 0;
|
|
130
|
-
|
|
131
|
-
logger.debug("Enter GoalItem.right_predict: " + this.id);
|
|
132
|
-
var items = chart.get_items_from(this.data.from);
|
|
133
|
-
items.forEach(function(item) {
|
|
134
|
-
if ((typeOf(item) === "doubledotteditem") && (item.data.right_dot < item.data.rule.rhs.length)){
|
|
135
|
-
if (grammar.is_headcorner_of(item.data.rule.lhs, that.data.nonterminal)) {
|
|
136
|
-
for (var i = item.data.to; i <= that.data.to; i++) {
|
|
137
|
-
for (var j = i; j <= that.data.to; j++) {
|
|
138
|
-
var new_goal = new GoalItem(i, j, item.data.rule.rhs[item.data.right_dot]);
|
|
139
|
-
nr_items_added += agenda.add_item(new_goal, chart);
|
|
140
|
-
logger.debug("GoalItem.right_predict: " + that.id +", " + item.id + " |- " + new_goal.id);
|
|
141
|
-
}
|
|
142
|
-
}
|
|
143
|
-
}
|
|
144
|
-
}
|
|
145
|
-
});
|
|
146
|
-
logger.debug("Exit GoalItem.right_predict: number of items added: " + nr_items_added);
|
|
147
|
-
return(nr_items_added);
|
|
148
|
-
};
|
|
149
|
-
|
|
150
|
-
GoalItem.prototype.left_complete = function (chart, agenda, grammar) {
|
|
151
|
-
var that = this;
|
|
152
|
-
var nr_items_added = 0;
|
|
153
|
-
|
|
154
|
-
logger.debug("Enter GoalItem.left_complete: " + this.id);
|
|
155
|
-
for (var j = this.data.from; j <= this.data.to; j++) {
|
|
156
|
-
var items1 = chart.get_items_from_to(this.data.from, j);
|
|
157
|
-
var items2 = chart.get_items_from_to(j, this.data.to);
|
|
158
|
-
items1.forEach(function(item1) {
|
|
159
|
-
if (typeOf(item1) === "cyk_item") {
|
|
160
|
-
items2.forEach(function(item2) {
|
|
161
|
-
if ((typeOf(item2) === "doubledotteditem") && (item2.data.left_dot > 0)) {
|
|
162
|
-
logger.debug("GoalItem.left_complete: trying: " + that.id + ", " + item2.id + ", " + item1.id);
|
|
163
|
-
if ((item2.data.rule.rhs[item2.data.left_dot-1] === item1.data.rule.lhs) &&
|
|
164
|
-
grammar.is_headcorner_of(item2.data.rule.lhs, that.data.nonterminal)) {
|
|
165
|
-
new_item = item2.copy();
|
|
166
|
-
new_item.recognise_left(item1);
|
|
167
|
-
nr_items_added += agenda.add_item(new_item, chart);
|
|
168
|
-
logger.debug("GoalItem.left_complete: " + that.id +", " + item2.id + ", " + item1.id + " |- " + new_item.id);
|
|
169
|
-
}
|
|
170
|
-
}
|
|
171
|
-
});
|
|
172
|
-
}
|
|
173
|
-
});
|
|
174
|
-
}
|
|
175
|
-
logger.debug("Exit GoalItem.left_complete: number of items added: " + nr_items_added);
|
|
176
|
-
return(nr_items_added);
|
|
177
|
-
};
|
|
178
|
-
|
|
179
|
-
GoalItem.prototype.right_complete = function (chart, agenda, grammar) {
|
|
180
|
-
var nr_items_added = 0;
|
|
181
|
-
var that = this;
|
|
182
|
-
|
|
183
|
-
logger.debug("Enter GoalItem.right_complete: " + this.id);
|
|
184
|
-
for (var j = this.data.from; j <= this.data.to; j++) {
|
|
185
|
-
var items1 = chart.get_items_from_to(this.data.from, j);
|
|
186
|
-
var items2 = chart.get_items_from_to(j, this.data.to);
|
|
187
|
-
items1.forEach(function(item1) {
|
|
188
|
-
if (typeOf(item1) === "doubledotteditem") {
|
|
189
|
-
items2.forEach(function(item2) {
|
|
190
|
-
if (typeOf(item2) === "cyk_item") {
|
|
191
|
-
logger.debug("GoalItem.right_complete: trying: " + that.id + ", " + item1.id + ", " + item2.id);
|
|
192
|
-
if ((item1.data.rule.rhs[item1.data.right_dot] === item2.data.rule.lhs) &&
|
|
193
|
-
grammar.is_headcorner_of(item1.data.rule.lhs, that.data.nonterminal)) {
|
|
194
|
-
new_item = item1.copy();
|
|
195
|
-
new_item.recognise_right(item2);
|
|
196
|
-
nr_items_added += agenda.add_item(new_item, chart);
|
|
197
|
-
logger.debug("GoalItem.right_complete: " + that.id +", " + item1.id + ", " + item2.id + " |- " + new_item.id);
|
|
198
|
-
}
|
|
199
|
-
}
|
|
200
|
-
});
|
|
201
|
-
}
|
|
202
|
-
});
|
|
203
|
-
}
|
|
204
|
-
logger.debug("Exit GoalItem.right_complete: number of items added: " + nr_items_added);
|
|
205
|
-
return(nr_items_added);
|
|
206
|
-
};
|